* Testing, testing, testing

* Allow limiting of which "keep state" rules generate netflow record
  - One way: lookup rule label and allow generation of flow records for
    states whose labels match
    - What to do if ruleset changes under us?

* NetFlow v.5 support
  - A few extra fields
  - Very easy

* NetFlow v.9 export
  - Supports IPv6!
  - http://www.cisco.com/warp/public/cc/pd/iosw/prodlit/tflow_wp.htm

* sflow export (www.sflow.org)

* IPv6 export support
  - Needs sflow or NetFlow v.9

* Cleanup of code
  - Kill globals in favour of ctxt

* Import features from softflowd
  - Basic statistics
  - Control program

* More netflow fields
  - Most require kernel support
  - May be able to (partially) infer tcp flags

* Companion daemon which reads from pflog to log blocked/dropped packets
  (maybe)

* Kernel support for reporting state counter wraparound
  - Check counters on kernel side
  - Upon receipt of a packet which would cause either flow counter to wrap,
    send a (new) pfsync info message with the state data
  - Reset state's counters to zero + len of current packet

* Kernel changes to account for generated packets
  - Which ones? ICMP is arguably not appropriate...

